Can serum and urinary biomarkers predict acute kidney injury after robotic partial nephrectomy? a prospective feasibility study

Aim of the study: Acute kidney injury (AKI) is a complex event after partial nephrectomy (PN), associated with new-onset chronic kidney disease (CKD), cardiovascular events and overall mortality. As such, early prediction of AKI would allow optimization of postoperative care. In this study, we evaluated serum and urinary biomarkers for early prediction of AKI after robotic PN. Materials and methods: We prospectively enrolled adult patients with a cT1N0M0 renal mass, contralateral functioning kidney, eGFR >45ml/min/1,73m2 undergoing robotic enucleation by a single surgeon from July 2017 to February 2018. We evaluated biomarkers of glomerular dysfunction and/or tubular damage before surgery and 4,12 and 24 h after PN: eGFR, sNGAL, sCystatinC (CysC), proteinuria, u [TIMP-2]/[IGFBP7] ratio (NephroCheck® AKIRisk score) and uNGAL. We obtained “kinetic eGFRs” (kGFR) as estimates of creatinine clearance (reflecting dynamic changes in renal function) from two sCreatinine measurements at different time points after PN (kGFRpreop-4h: kGFR preop-12h; kGFR preop-24h; kGFR4h-12h; kGFR4h-24h; kGFR12h-24h). The study endpoint was AKI. ROC curve and univariate logistic regression analysis assessed the ability of biomarkers to predict AKI. Results: Overall, 40 patients were enrolled. Of these, 16 (40%) experienced AKI. Preoperative patient and tumor characteristics were comparable between the study groups. Mean post-operative change (Δ) CysC levels were significantly higher in patients who experienced AKI at all time points as compared to those who did not (4h p = 0,019; 12h 0,014; C 24h p=<0,0001). The ROC curve analysis confirmed these findings: Δ CysC levels showed good predictive ability for detection of AKI at 24h (AUC 0.84, p = 0.0001), 4h and 12h (AUC 0.74 and 0.73, both p = 0.01). Of note, we recorded an increase in kGFR from baseline at all time points in both study groups; yet, patients experiencing AKI showed a significantly lower %change in kGFR (preop-12h p=0,002; preop-24h p<0.001; 12h-24h p=0.002; 4h- 24h p = 0.002).At the ROC curve analysis, the highest predictive ability for detection of AKI was achieved by the %change in kGFR 24h after PN from baseline (AUC 0.86, p < 0.0001). At univariable analysis, only Δ CysC levels at all time points (4h OR: 2.21, p = 0.03, 12h OR:2.09 p = 0.0037; 24h OR: 4.09 p = 0.005) and %change in kinetic eGFR from baseline( preop-12h OR 0.90 p = 0.008, preop-24h OR 0.82 p = 0,001,4h-12h OR:0.95 p=0.001,4h-24h OR: 0.92 p=0.004;12h-24h OR: 0.90 p = 0.003) were shown to be significant predictors of AKI. Postoperative changes in sCysC levels accurately predicted the occurrence of AKI at all time points, particularly at 24 h. The % change in kinetic GFR from baseline, which may quantitate the dynamic variation of eGFR in response to PN “injury”, was significantly higher in patients not experiencing AKI at almost all time points. Discussion: These biomarkers appear promising for early (24h) or even very early (4–12h) prediction of AKI after PN, toward a concept of personalized risk-adapted postoperative care.